Predicting viral load suppression by self-reported adherence, pharmacy refill counts and real time medication monitoring among people living with HIV in Tanzania.

INTRODUCTION: Monitoring of adherence to antiretroviral treatment (ART) is of utmost importance to prevent treatment failure. Several measures to monitor adherence have been applied in low-resource settings and they all have pros and cons. Our objective was to examine whether any of the following adherence measures is a better predictor of participants' viral load suppression: (1) self-report, (2) pharmacy refill count, (3) Real Time Medication Monitoring (RTMM), (4) a combination of self-report and pharmacy refill count or (5) all three adherence assessment methods combined. METHODOLOGY: This was a post-hoc analysis of data from our 48-week REMIND-HIV randomized controlled trial in which adherence to ART was measured using self-report, pharmacy refill counts and RTMM among ART-experienced adults living with HIV subjectively judged to be nonadherent to ART. For each adherence measure, we calculated sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) for predicting virological failure defined as a viral load (VL) of > 20 copies/mL. To determine at which percentage of adherence the prediction was strongest, we evaluated adherence cut-offs of 80%, 85%, 90%, 95% and 100% using receiver operating characteristic (ROC) curves. VL data were obtained after 48 weeks of follow-up in the trial. RESULTS: A total of 233 people living with HIV (PLHIV) were included in this analysis. When comparing the ability of self-reported adherence with pharmacy refill count and RTMM adherence to predict viral load > 20 copies/ml, self-reported adherence had the lowest sensitivity, ranging from 6 to 17%, but the highest specificity, ranging from 100 to 86%, depending on cut-off values from 80 to 100%. Area under the ROC curves (AUC) were 0.54 for RTMM, 0.56 for pharmacy refill count and 0.52 for self-report, indicating low discriminatory capacity for each of the adherence measures. When we combined the self-report and pharmacy refill count measures, sensitivity increased, ranging from 28 to 57% but specificity decreased, ranging from 83 to 53%. When all three measures were combined, we observed the highest value of sensitivity, ranging from 46 to 92%, and PPV, ranging from 32 to 36%, at high cut-offs ranging from 80 to 100%. Upon combination of three adherence measures, the AUC increased to 0.59. CONCLUSION: Our results show that adherence assessed exclusively by self-report, pharmacy refill count or RTMM were insufficiently sensitive to predict virologic failure. Sensitivity markedly improved by combining all three measures, but the practical feasibility of such an approach would need to be studied.

Feasibility of SMS to remind pregnant and breastfeeding women living with HIV to take antiretroviral treatment in Kilimanjaro region, Tanzania: a pilot study.

BACKGROUND: Pregnant and breastfeeding Women Living with HIV (WLHIV) often have difficulties in reaching adequate levels of adherence (>95%) to Antiretroviral treatment. "Forgetting" is the most commonly mentioned reason. Sending reminders via SMS is expected to improve adherence. We conducted a pilot study to investigate acceptability, user experience and technical feasibility of sending reminder-SMS to WLHIV. METHODS: This was a 6-months observational pilot-study among WLHIV attending antenatal and postnatal care at Kilimanjaro Christian Medical Centre in Moshi, Tanzania. Women received a reminder-SMS 30 minutes before usual time of intake. One hour later, they received an SMS asking whether they took medication to which they could reply with 'Yes' or 'No'. Messages were sent 3 times a week on randomly chosen days to prevent reliance on daily messages. We calculated the percentage of number of SMS delivered, failed to be delivered, and replied to. We analysed feedback from exit-interviews about experience with the SMS-reminders. RESULTS: 25 women were enrolled (age 18-45), 2 were lost to follow up. 5,054 messages were sent of which 53 failed to be delivered (1%). 1,880 SMS were sent with a question if medication was taken; 1,012 (54%) messages were replied to, of which 1,003 (99%) were replied with 'YES' and closely to 'YES', and a total of 9 (1%) with 'NO' and 'closely to NO'. 868 messages (46%) were not responded to due to either dropout, change of phone number, loss of phone or network failure. Results from 18 interviews showed that 16 (89%) women were satisfied with SMS reminders. 2 (11%) were concerned about unwanted disclosure because of the content 'don't forget to take medication' and one reported other privacy issues (6%). 3 (17%) women experienced stigma. CONCLUSION: 99%of SMS being delivered indicates that SMS reminders in this resource-limited setting are technically feasible. However, concerns regarding privacy were noted, specifically the risk of unwanted disclosure and the experience of stigma. Participants indicated that being made aware of their adherence, motivated them to adhere better. However, personalised and more neutral content of the SMS might be a way to improving the intervention.

Comparison of overnight pooled and standard sputum collection method for patients with suspected pulmonary tuberculosis in northern Tanzania.

In Tanzania sputum culture for tuberculosis (TB) is resource intensive and available only at zonal facilities. In this study overnight pooled sputum collection technique was compared with standard spot morning collection among pulmonary TB suspects at Kibong'oto National TB Hospital in Tanzania. A spot sputum specimen performed at enrollment, an overnight pooled sputum, and single morning specimen were collected from 50 subjects and analyzed for quality, quantity, and time to detection in Bactec MGIT system. Forty-six (92%) subjects' overnight pooled specimens had a volume ≥5 mls compared to 37 (37%) for the combination of spot and single morning specimens (P < 0.001). Median time to detection was 96 hours (IQR 87-131) for the overnight pooled specimens compared to 110.5 hours (IQR is 137 right 137-180) for the combination of both spot and single morning specimens (P = 0.001). In our setting of limited TB culture capacity, we recommend a single pooled sputum to maximize yield and speed time to diagnosis.